Page 1 of 93
ARTIBA ARTIFICIAL INTELLIGENCE ENGINEER CERTIFICATION
EXAM COMPLETE QUESTIONS AND CORRECT ANSWERS
LATEST UPDATE THIS YEAR
SHORT POINT FORM COVERAGE
• AI = ML + DL + data + deployment
• ML types = supervised, unsupervised, reinforcement
• Preprocessing = clean, encode, scale
• Algorithms = regression, trees, clustering
• Evaluation = precision, recall, F1
• DL = neural networks, CNN, RNN
• NLP = text processing + embeddings
• Deployment = API + cloud + monitoring
• Ethics = bias, fairness, privacy
• Business = ROI + real-world use
, Page 2 of 93
MCQs WITH RATIONALES (BATCH 1: 50 QUESTIONS)
Question 1
A data scientist is building a model to predict housing prices using labeled historical data. Which
type of machine learning is being applied?
A. Unsupervised learning
B. Reinforcement learning
C. Supervised learning
D. Deep learning only
Answer: C
Rationale: Labeled data with known outputs indicates supervised learning.
Question 2
A dataset contains missing numerical values in several columns. What is the most appropriate
preprocessing step?
A. Remove entire dataset
B. Ignore missing values
C. Impute using mean or median values
D. Convert to categorical
, Page 3 of 93
Answer: C
Rationale: Imputation preserves data while handling missing values.
Question 3
A model performs well on training data but poorly on unseen data. What issue is most likely
occurring?
A. Underfitting
B. Overfitting
C. High bias
D. Data leakage
Answer: B
Rationale: Overfitting occurs when a model memorizes training data but fails to generalize.
Question 4
Which algorithm is best suited for classification problems involving nonlinear decision
boundaries?
A. Linear regression
B. k-Means clustering
C. Support Vector Machine
D. PCA
, Page 4 of 93
Answer: C
Rationale: SVM handles complex boundaries using kernels.
Question 5
A confusion matrix shows high false positives. Which metric is most affected?
A. Recall
B. Precision
C. Accuracy
D. F1-score
Answer: B
Rationale: Precision decreases when false positives increase.
Question 6
A neural network uses ReLU activation in hidden layers. What is its primary advantage?
A. Outputs probabilities
B. Prevents vanishing gradients
C. Reduces data size
D. Handles categorical data
ARTIBA ARTIFICIAL INTELLIGENCE ENGINEER CERTIFICATION
EXAM COMPLETE QUESTIONS AND CORRECT ANSWERS
LATEST UPDATE THIS YEAR
SHORT POINT FORM COVERAGE
• AI = ML + DL + data + deployment
• ML types = supervised, unsupervised, reinforcement
• Preprocessing = clean, encode, scale
• Algorithms = regression, trees, clustering
• Evaluation = precision, recall, F1
• DL = neural networks, CNN, RNN
• NLP = text processing + embeddings
• Deployment = API + cloud + monitoring
• Ethics = bias, fairness, privacy
• Business = ROI + real-world use
, Page 2 of 93
MCQs WITH RATIONALES (BATCH 1: 50 QUESTIONS)
Question 1
A data scientist is building a model to predict housing prices using labeled historical data. Which
type of machine learning is being applied?
A. Unsupervised learning
B. Reinforcement learning
C. Supervised learning
D. Deep learning only
Answer: C
Rationale: Labeled data with known outputs indicates supervised learning.
Question 2
A dataset contains missing numerical values in several columns. What is the most appropriate
preprocessing step?
A. Remove entire dataset
B. Ignore missing values
C. Impute using mean or median values
D. Convert to categorical
, Page 3 of 93
Answer: C
Rationale: Imputation preserves data while handling missing values.
Question 3
A model performs well on training data but poorly on unseen data. What issue is most likely
occurring?
A. Underfitting
B. Overfitting
C. High bias
D. Data leakage
Answer: B
Rationale: Overfitting occurs when a model memorizes training data but fails to generalize.
Question 4
Which algorithm is best suited for classification problems involving nonlinear decision
boundaries?
A. Linear regression
B. k-Means clustering
C. Support Vector Machine
D. PCA
, Page 4 of 93
Answer: C
Rationale: SVM handles complex boundaries using kernels.
Question 5
A confusion matrix shows high false positives. Which metric is most affected?
A. Recall
B. Precision
C. Accuracy
D. F1-score
Answer: B
Rationale: Precision decreases when false positives increase.
Question 6
A neural network uses ReLU activation in hidden layers. What is its primary advantage?
A. Outputs probabilities
B. Prevents vanishing gradients
C. Reduces data size
D. Handles categorical data
